html5-img
1 / 17

Chapter 6 - Specific Case Study: Searching for Bent-double Galaxies

Chapter 6 - Specific Case Study: Searching for Bent-double Galaxies . Chandrika Kamath Center for Applied Scientific Computing Lawrence Livermore National Laboratory http://www.llnl.gov/casc/people/kamath. UCRL-PRES-145087: The work of Chandrika Kamath in Chapters 5, 6, and 7

garson
Download Presentation

Chapter 6 - Specific Case Study: Searching for Bent-double Galaxies

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Chapter 6 - Specific Case Study: Searching for Bent-double Galaxies Chandrika Kamath Center for Applied Scientific Computing Lawrence Livermore National Laboratory http://www.llnl.gov/casc/people/kamath UCRL-PRES-145087: The work of Chandrika Kamath in Chapters 5, 6, and 7 was performed under the auspices of the U.S. Department of Energy by the University of California Lawrence Livermore National Laboratory under contract No. W-7405-Eng-48.

  2. Mining the FIRST survey for galaxies with a bent-double morphology • FIRST: Faint Images of the Radio Sky at Twenty Centimeters • Radio equivalent of the Palomar Observatory Sky Survey (POSS) • 10,000 square degrees survey of the North Galactic Cap • Using the NRAO Very Large Array (VLA), B configuration

  3. The FIRST data • 1.8’’ pixels, resolution 5’’, rms 0.15mJy • ~90 radio sources per square-degree at 1mJy threshold • The morphological type of a radio source provides clues to their emission mechanism, source class, and the properties of the surrounding medium • The raw data from the telescopes is extensively processed • Images maps and catalog available (sundog.stsci.edu)

  4. Radio-emitting galaxies (radio-sources) exhibit a wide range of morphological types

  5. We are using data mining to find “bent-doubles” in FIRST • FIRST astronomers interested in “bent-doubles” • indicates presence of clusters of galaxies • first “identify” using a visual technique • followed by optical observations and checks with other surveys • Visual identification is no longer feasible • subjective, tedious, likely to miss cases …. • ~900,000 galaxies in the full survey  Our goal is to replace the visual identification of bent doubles by a semi-automated one. Details at http://www.llnl.gov/casc/sapphire

  6. Image Map 1150 pixels Catalog 720K entries 1550 pixels ~32K image maps, 7.1MB each 64 pixels Detecting bent-double galaxies in 250GB image data, 78MB catalog data (as of 7/00)

  7. Our approach for finding bent doubles in the FIRST data • Group the catalog entries into a “galaxy” • Separate sources based on number of catalog entries • 1-entry sources unlikely to be bent-doubles • > 3-entry sources all “interesting” • study the 2- and 3-entry sources separately • results in splitting a small training set (313 ---> 118 + 195)

  8. Extract Derived Features Create Decision Tree Check for Accuracy Catalog Extract Features for Unclassified Data Training Set Update Training Set Show Results, Obtain Score Apply Decision Tree to Unclassified Data Our approach for finding bent doubles in the FIRST data • Calculate features for a galaxy (103 features) • Use the features to train a decision tree • Use the tree to classify the unlabeled galaxies and validate the results • Use validated results to enhance training set

  9. Potential features used to describe a radio galaxy must be carefully selected • Features must be • relevant to the problem • robust • scale, rotation, translation invariant • Feature extraction is an iterative process • Feature extraction requires • input from the domain scientists • an understanding on how the data was collected

  10. Potential features that can be used to describe a radio galaxy • Single, for each CE • peak (integrated) flux, area, major (minor) axis, ellipticity (major/minor), position angle (PA) …. • Double, for CEs taken 2-at-a-time • maximum flux, relative flux (ratio of the fluxes), total flux, total area, distance between centers, angle between major axes …. • Triple, for CEs taken 3-at-a-time • maximum flux, total flux, total area, distances, angles ….

  11. N Calculating robust pair-wise angles is difficult • Position angle differences are sensitive to orientation • Need to determine angles geometrically • What is the position angle for point sources? • {point, point}: set angle to • {point, extended}: set angle to A A B B M M

  12. Our Python-based tool allows validation of results and incorporation of user feedback

  13. Our initial results using a single tree for 3-entry sources were satisfactory • Labeled training set: 167 bents, 28 non-bents • Performed several inner iterations using pruned trees (c5.0 decision tree software) • Ten 10-fold cross-validation errors: mean(SE) • using all the features: 9.7%(0.3%) • using triple features only: 10.7%(0.3%) • Discriminating features include geometrically calculated angles, relative distances, ellipticity and symmetry measures A bent-double missed in a manual search:

  14. Accurate classification of 2-entry sources needs more refinement • Labeled training set: 72 bents, 46 non-bents • Decision trees depend on the training/test set • Ten 10-fold cross-validation errors: 19.8%(1.0%) • Reasons why classification is difficult • small labeled set • lose information in going from images to catalog • irrelevant features Bent-double Non-bent-double

  15. Our results with ensembles indicate a performance improvement • Very small ensembles have higher error • Bagging is slightly less accurate than other methods • Very large ensembles may be wasteful % error (SE) using all features

  16. General observations from mining the FIRST survey • Easy (on-line) access to data is a big help • Well-supported public domain tools for reading, writing, and viewing data are very useful • Availability of the catalog allowed a head-start on the classification of the galaxies • Generating a “good” training set is non-trivial • few bent- and non-bent- doubles identified in a visual inspection • disagreement among astronomers on cases which are difficult to classify • lack of ground truth • Extracting relevant features in a robust manner is difficult

  17. Chapter 6 - References FIRST web page: sundog.stsci.edu. Includes papers, access to the catalog, and image cutouts Kamath, C., E. Cantú-Paz, I. K. Fodor, and N. Tang, “Searching for bent-double galaxies in the first survey”, in Data Mining for Scientific and Engineering Applications, R. Grossman, C. Kamath, W. Kegelmeyer, V. Kumar, and R. Namburu (eds.), Kluwer 2001. Sapphire project web page at http://www.llnl.gov/casc/sapphire

More Related